This approach relies on (i) a supervised convolutional autoencoder architecture that incorporates the temporal ... 3)] 0 masking_1 (Masking) (None, None, 3) 0 conv1d_2 (Conv1D) (None, None, 64) 1600 ...